I built an AI Literacy course
How I expanded my capstone into a four-module, research-grounded AI literacy program on a custom platform in roughly 210 hours, and what that compression says about the future of learning and development.
ritchot.me
Writing by Michael Ritchot.
Latest posts
reflections, and the state of LLMs at the end of 2025
A year-end look at LLMs becoming ordinary software: world models, the autonomy slider, the local-compute wall, and why 2026 is for building instead of benchmarking.
GPT-5 has come a long way in Mathematics
Re-running last year's CEMC test against GPT-5: 98% per-attempt accuracy ends the era of unreliable AI mathematics, and institutional responses have not kept pace.
on writing, and an MIT study
The MIT 'Your Brain on ChatGPT' study is being misread: what it actually shows, where its methodology wobbles, and why the real indictment is of timed essay assessment.
some thoughts on emergent technology and the future of education
Emerging technologies will restructure labor faster than schools are adapting: a case for earlier specialization, bolder curricula, and taking AI-driven disruption of teaching seriously.
o1 pro mode still has a long way to go for Mathematics
Testing OpenAI's o1 pro mode on 60 University of Waterloo Problems of the Week: a 67% pass rate, with image interpretation the clear weak point, means my mathematics classroom is still safe.
AI: Knowing The Gods We Have Created
From Deus Ex's Morpheus to AlphaGo, OpenAI Five, and ChatGPT: how AI moved from fiction to household fixture, and what it means to create gods we may end up following.
From Private to Shared Knowledge
After years of keeping ideas in notebooks, I built a place to think in public, at my own pace, without the content treadmill.